A Video-KYC flow live in nine days

A wealth and investment app faces a different onboarding pressure than a lender: the account often has to be usable quickly to capture intent, but the identity check gates money movement and sits under investor-onboarding scrutiny. The operational problem is standing up a Video-KYC-style flow that combines ID-OCR, face match and liveness, and doing it quickly enough to launch, without shipping a flow whose error behaviour nobody has characterised.

The useful lens here is that two numbers describe the health of such a flow: how often it is right, and how often it finishes without a human. Accuracy tells you the assurance floor; straight-through processing tells you the operating cost and the size of the manual tail. Neither number alone is sufficient, and a fast go-live means little if the audit trail behind these numbers is thin.

HyperVerge, the publisher, reports that IndMoney moved from concept to go-live in 9 days using a digital Video-KYC flow that combines ID-OCR, face match and liveness for automated identity verification. HyperVerge states the deployment was associated with roughly 4x business-volume growth, an approximately 99.5% verification accuracy figure, and about 80% straight-through processing.

As with the other HyperVerge material, no publication date is stated. The accuracy and STP figures are therefore undated vendor-reported claims without a defined measurement window or definition of what counts as a correct verification or a straight-through case, and an evaluator should treat them as starting questions rather than settled benchmarks and re-verify them directly.

An SMB should pin down definitions before trusting the headline numbers. Ask what '99.5% accuracy' is measured against: which population, which document types, over what period, and whether it counts false accepts and false rejects the same way. Ask what the ~80% straight-through figure excludes: is the remaining fifth a manual queue, a hard reject, or a retry loop that frustrates genuine users? A 9-day go-live is impressive but says nothing about these definitions.

Inspect what the audit log captures per decision, because in this kind of flow the log is the product. It should let you reconstruct any verification end to end: the OCR fields extracted, the face-match and liveness scores, the consent captured at the moment of collection, and any human override, all time-stamped. Confirm applicable rules, consent, retention and escalation design with qualified counsel or compliance owners.

The 4x growth and the STP share are properties of IndMoney's specific applicant mix, document quality and risk tolerance. An SMB with a different regional ID profile, lower image quality, or a customer base less comfortable with a live selfie step will not reproduce the same straight-through rate, and pushing thresholds to match it trades directly against the accuracy figure. The two numbers are coupled, so importing one target without the other misleads.

The onboarding context is an investment app, whose identity requirements are specific to its product and licensing. Whether any such obligations bind a given SMB, and which verification mechanism is appropriate for its product, are structural questions of scope; confirm applicable rules, consent, retention and escalation design with qualified counsel or compliance owners rather than assume they carry over from a wealth platform's setup.